Pubblicazioni

Using the basic GenBank local alignment search tool program (BLAST) to identify fungi collected in a recently protected beech forest at Montricher (Switzerland), the number of ITS sequences associated to the wrong taxon name appears to be around 30%, even higher than previously estimated. Such results rely on the in-depth re-examination of BLAST results for the most interesting species that were collected, viz. first records for Switzerland, rare or patrimonial species and problematic species (when BLAST top scores were equally high for different species), all belonging to Agaricomycotina. This paper dissects for the first time a number of sequence-based identifications, thereby showing in every detail—particularly to the user community of taxonomic information—why sequence-based identification in the context of a fungal inventory can easily go wrong. Our first conclusion is that in-depth examination of BLAST results is too time consuming to be considered as a routine approach for future inventories: we spent two months on verification of approx. 20 identifications. Apart from the fact that poor taxon coverage in public depositories remains the principal impediment for successful species identification, it can be deplored that even very recent fungal sequence deposits in GenBank involve an uncomfortably high number of misidentifications or errors with associated metadata. While checking the original publications associated with top score sequences for the few examples that were here re-examined, a positive consequence is that we uncovered over 80 type sequences that were not annotated as types in GenBank. Advantages and pitfalls of sequence-based identification are discussed, particularly in the light of undertaking fungal inventories. Recommendations are made to avoid or reduce some of the major problems with sequence-based identification. Nevertheless, the prospects for a more reliable sequence-based identification of fungi remain quite dim, unless authors are ready to check and update the metadata associated with previously deposited sequences in their publications.